Abstract
Federated learning is an increasingly popular paradigm that enables a large number of entities to collaboratively learn better models. In this work, we study minimax group fairness in federated learning scenarios where different participating entities may only have access to a subset of the population groups during the training phase. We formally analyze how our proposed group fairness objective differs from existing federated learning fairness criteria that impose similar performance across participants instead of demographic groups. We provide an optimization algorithm – FedMinMax – for solving the proposed problem that provably enjoys the performance guarantees of centralized learning algorithms. We experimentally compare the proposed approach against other state-of-the-art methods in terms of group fairness in various federated learning setups, showing that our approach exhibits competitive or superior performance.
Type
Publication
In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency (FAccT 2022)
We propose FedMinMax, an algorithm for achieving minimax demographic group fairness in federated learning settings where different clients may only have access to a subset of population groups during training. Unlike existing federated fairness criteria that enforce similar performance across participants, FedMinMax targets fairness across demographic groups directly.
A key contribution is the formal analysis showing how our group fairness objective differs from client-level fairness, along with conditions under which they coincide. FedMinMax provably enjoys the same performance guarantees as centralized learning algorithms, and our experiments across multiple federated setups confirm that it achieves competitive or superior group fairness compared to state-of-the-art methods.
Image credit: paper